Re: When is %EOF not an %EOF

[Simon Coulter <shc@xxxxxxxxxxxxxxxxx>]

On Tuesday, September 30, 2003, at 12:19  AM, DeLong, Eric wrote:

> I find it very interesting that by simply changing a reade to a read 
> and then testing the keys manually can improve performance so 
> dramatically.

The secret is blocking. Fixing the I/O is the single best thing you can 
do to fix performance problems (presuming there are no physical 
constraints like insufficient main storage, or a gutless processor).

As an example I changed a conversion process at a previous employer to 
reorganise the work files into keyed sequential order, used a large 
blocking factor, and dropped the run time from 3 days to 12 hours.

I wouldn't suggest that you religiously use "READ and test" in place of 
READE especially if the values you want are widely separated (i.e., a 
sparse key) but whenever you are processing large amounts of data 
sequentially by key a good old-fashioned sort followed by sequential 
reads can significantly improve the run time.

Regards,
Simon Coulter.
--------------------------------------------------------------------
   FlyByNight Software         AS/400 Technical Specialists

   http://www.flybynight.com.au/
   Phone: +61 3 9419 0175   Mobile: +61 0411 091 400        /"\
   Fax:   +61 3 9419 0175   mailto: shc@xxxxxxxxxxxxxxxxx   \ /
                                                             X
                 ASCII Ribbon campaign against HTML E-Mail  / \
--------------------------------------------------------------------